Intellect as Distinct From Openness: Differences Revealed ...

Journal of Personality and Social Psychology © 2009 American Psychological Association
2009, Vol. 97, No. 5, 883– 892 0022-3514/09/$12.00 DOI: 10.1037/a0016615

Intellect as Distinct From Openness: Differences Revealed by fMRI of
Working Memory

Colin G. DeYoung Noah A. Shamosh and Adam E. Green

University of Minnesota Yale University

Todd S. Braver Jeremy R. Gray

Washington University Yale University

Relatively little is known about the neural bases of the Big Five personality trait Openness/Intellect. This
trait is composed of 2 related but separable aspects, Openness to Experience and Intellect. On the basis
of previous behavioral research (C. G. DeYoung, J. B. Peterson, & D. M. Higgins, 2005), the authors
hypothesized that brain activity supporting working memory (WM) would be related to Intellect but not
to Openness. To test this hypothesis, the authors used functional magnetic resonance imaging (fMRI) to
scan a sample of 104 healthy adults as they performed a difficult WM task. Intellect (and not Openness)
was found to correlate with WM accuracy and with accuracy-related brain activity, in left lateral anterior
prefrontal cortex and posterior medial frontal cortex. Neural activity in these regions mediated the
association between Intellect and WM performance, implicating these regions in the neural substrate of
Intellect. Intellect was also correlated significantly with scores on tests of intelligence and WM capacity,
but the association of Intellect with brain activity could not be entirely explained by cognitive ability.

Keywords: intellect, openness, prefrontal cortex, working memory, intelligence

The five-factor model or Big Five classifies personality traits dite, clever, etc.1 There are also many terms (though perhaps
into five broad domains: Extraversion, Agreeableness, Conscien- fewer) that describe Openness to Experience: artistic, perceptive,
tiousness, Neuroticism, and Openness/Intellect. The compound poetic, fantasy-prone, etc. Additionally, there are many terms that
label for the last of these traits reflects an old debate about how could characterize people high in Intellect or Openness or both:
best to characterize the content of this domain, with some research- imaginative, original, complex, innovative, etc. (Indeed, Saucier,
ers preferring “Openness to Experience” (e.g., Costa & McCrae, 1992, proposed that “imagination” might be a better single label
1992a) and others preferring “Intellect” (e.g., Goldberg, 1993). for the domain as a whole.) Johnson (1994) demonstrated that, of
This debate has been largely resolved conceptually by the obser- the facets (subtraits) in the Revised NEO Personality Inventory
vation that “Openness” and “Intellect” describe two related but (NEO-PI-R; Costa & McCrae, 1992b), the Ideas and Aesthetics
separable aspects of the larger domain (Johnson, 1994; Saucier, facets were the purest markers of the lexical Openness/Intellect
1992). Lexical studies make it clear that both aspects are well factor, and he suggested that interests in truth and beauty are
represented in natural language and that content related to both complementary qualities at the heart of the Openness/Intellect
appears among the terms loading on a single Big Five factor (e.g., domain. McCrae and Costa (1997) argued that this domain reflects
Goldberg, 1990; Saucier, 1992). There are many terms in English the “breadth, depth, and permeability of consciousness” (p. 826).
that describe Intellect: intellectual, intelligent, philosophical, eru- Perhaps Openness reflects these qualities of consciousness in
relation to sensory or perceptual information, whereas Intellect
Colin G. DeYoung, Department of Psychology, University of Minne- reflects them in relation to abstract or semantic information.
sota; Noah A. Shamosh and Adam E. Green, Department of Psychology,
Yale University; Todd S. Braver, Department of Psychology, Washington The identification of Openness and Intellect as the two major
University; Jeremy R. Gray, Department of Psychology and Interdepart- aspects of this domain of personality was recently given empirical
mental Neuroscience Program, Yale University. support by a factor analysis of 15 scales measuring facets of
Openness/Intellect (DeYoung, Quilty, & Peterson, 2007). The
This research was supported by National Institute of Mental Health covariance of those facets indicated the presence of two correlated
Grants MH R01 6608, awarded to Jeremy R. Gray, and F32 MH077382, factors, clearly recognizable as Openness and Intellect. Each factor
awarded to Colin G. DeYoung. The content is solely the responsibility of was strongly marked by six facets, suggesting their similarity in
the authors and does not necessarily represent the official views of the
National Institute of Mental Health or the National Institutes of Health. 1 Similar collections of terms have defined the Openness/Intellect factor
in other languages, except when terms reflecting intellectual ability have
Correspondence concerning this article should be addressed to Colin G. been omitted (e.g., studies in Dutch and Italian), in which case the factor
DeYoung, Department of Psychology, University of Minnesota, 75 East tends to tilt more toward content related to unconventionality (John,
River Road, Minneapolis, MN 55455 or to Jeremy R. Gray, Department of Naumann, & Soto, 2008).
Psychology, Yale University, Box 208205, New Haven, CT 06520.
E-mail: [email protected] or [email protected]

883

884 DEYOUNG, SHAMOSH, GREEN, BRAVER, AND GRAY

importance to the larger domain. The two factors were further lateral aPFC in abstract integration, as distinct from more basic
characterized by examining their correlations with over 2,000 WM processes such as maintenance and manipulation of informa-
items of the International Personality Item Pool (Goldberg, 1999). tion (Reynolds, West, & Braver, 2008). For example, activity in
This analysis revealed that Intellect encompasses traits reflecting left aPFC has been associated with abstract, relational integration in
intellectual engagement and perceived intelligence (e.g., “Avoid analogical reasoning (Green et al., 2006). Left aPFC appears similarly
philosophical discussions” (reversed); “Am quick to understand important for integration during mathematical problem solving (De
things”), whereas Openness encompasses traits reflecting artistic Pisapia, Slomski, & Braver, 2007), matrix reasoning (Christoff et
and contemplative qualities related to engagement in sensation and al., 2001), and episodic memory (Reynolds, McDermott, &
perception (e.g., “Believe in the importance of art”; “See beauty in Braver, 2006). In a previous study of the sample examined here
things that others might not notice”; DeYoung et al., 2007). Hav- (Shamosh et al., 2008), we found that activity in left lateral aPFC
ing established the existence of these two related but separable was associated with individual differences in WM, intelligence,
aspects of Openness/Intellect, one important concern is discrimi- and the tendency to prefer larger, delayed rewards over smaller,
nant validity. How do these two aspects differ from each other in immediate rewards. The association of left lateral aPFC with WM
their associations with other variables? and intelligence suggests the hypothesis that this brain region is
associated with the trait of Intellect.
Levels of personality organization below the Big Five are asso-
ciated with unique genetic variance (Jang et al., 2002; Jang, The pMFC region is of interest because it is reliably engaged in
McCrae, Angleitner, Riemann, & Livesley, 1998), suggesting that WM tasks (Cabeza & Nyberg, 2000; Owen, McMillan, Laird, &
it may be possible to identify biological systems that differentiate Bullmore, 2005) and appears to be involved in the cognitive
Intellect from Openness. In the present study, we tested the hy- functions of monitoring performance during goal-directed activity
pothesis that brain function associated with working memory (Ridderinkhof, Ullsperger, Crone, & Nieuwenhuis, 2004) and de-
(WM) performance would be related to Intellect but not to Open- tecting the likelihood of error (Brown & Braver, 2005). Monitor-
ness. Intellectual individuals seem more likely than those who are ing one’s performance during cognitive tasks seems likely to be
simply open to experience to use brain systems that allow for associated with Intellect, especially given the degree to which
successful processing of information during WM tasks with a high Intellect reflects the tendency to be motivated and engaged by
cognitive demand. This hypothesis was derived in part from be- intellectual activities. We therefore hypothesized that this region
havioral research relating facets of Openness/Intellect to a battery might be among the neural correlates of Intellect. Finally, other
of tests of WM and other cognitive functions associated with parts of the canonical WM network (Cabeza & Nyberg, 2000;
prefrontal cortex (PFC) (DeYoung, Peterson, Higgins, 2005). In Owen et al., 2005; Wager & Smith, 2003), such as the dorsolateral
that study, like the present one, the Big Five was measured using PFC and parietal cortex, which have also been associated with tests
the NEO-PI-R (Costa & McCrae, 1992b), which divides each of intelligence (Gray & Thompson, 2004), may also be associated
of the Big Five into six facets. Four of these facets (Fantasy, with Intellect, though our predictions regarding these regions were
Aesthetics, Feelings, and Actions) clearly mark the Openness less specific than those for left aPFC and pMFC.
aspect of the domain in factor analysis (DeYoung et al., 2007). The
Ideas facet, however, is a good marker of Intellect (DeYoung et al., A key question in testing our hypotheses is whether any association
2007) and was the facet most strongly associated with performance of Intellect with brain function is due to the association of both
on the WM battery (DeYoung et al., 2005).2 The sixth facet, Intellect and WM with intelligence. The Ideas facet of the NEO-PI-R,
Values, does not mark either aspect strongly but was also found to our marker for Intellect, is the facet most consistently associated with
be associated with WM. scores on intelligence tests (DeYoung et al., 2005; Furnham, Dissou,
Sloan, & Chamorro-Premuzic, 2007; McCrae, 1993; Moutafi, Furn-
We used functional magnetic resonance imaging (fMRI) of ham, & Crump, 2003, 2006). Although WM is not identical to
individuals performing a difficult WM task in the present study in intelligence, the two are strongly related (Conway, Kane, & Engle,
order to test the hypothesis that Intellect (but not Openness) would 2003), suggesting that WM forms a key part of the cognitive substrate
be associated with WM performance and with brain activity sup- of intelligence. Additionally, the neural substrates of intelligence and
porting WM performance. Use of a large sample (by the standards WM are at least partially overlapping (Gray, Chabris, & Braver, 2003;
of neuroimaging research) enabled the identification of brain re- Gray & Thompson, 2004; Kane & Engle, 2002).
gions in which individual variation in neural activity predicted
WM performance. Because this analytic approach identifies re- If Intellect, like intelligence, is associated with the neural sub-
gions with meaningful variability, it appears particularly promising strates of WM, what will that finding add to our knowledge? We
for personality neuroscience (DeYoung & Gray, in press), in addressed this question empirically, using ability tests of intelli-
contrast to approaches that identify regions consistently activated
in the sample as a whole. 2 The fact that the NEO-PI-R contains only one clear Intellect facet and
four clear Openness facets is due to the inventory’s history and does not
Two brain regions of particular interest in our analyses were the constitute evidence that Intellect is peripheral to the larger Openness/
left lateral region of anterior PFC (aPFC; also called frontopolar Intellect domain. The facets of the NEO-PI-R were derived rationally;
cortex) and the region of posterior medial frontal cortex (pMFC) those for Openness to Experience were developed prior to McCrae and
that encompasses the dorsal anterior cingulate cortex (ACC). Left Costa’s (1985) attempt to harmonize the NEO with the lexical Big Five.
aPFC appears to support the abstract integration of information McCrae and Costa have consistently resisted the idea that Intellect might
from multiple cognitive operations (Gilbert et al., 2006; Green, be a valid interpretation of content in this domain (e.g., McCrae & Costa,
Fugelsang, Kraemer, Shamosh, & Dunbar, 2006; Ramnani & 1997). As noted above, however, considerable evidence in both lexical and
Owen, 2004) and has been implicated in intelligence (Jung & questionnaire research indicates that Intellect is just as central to the larger
Haier, 2007). Several neuroimaging studies have implicated left domain as is Openness.

INTELLECT AND fMRI OF WORKING MEMORY 885

gence. As noted above, measures of Intellect reflect perceived (59 women), ranging in age from 18 to 40 years (M ϭ 22.67, SD ϭ
intelligence as well as intellectual engagement. Neither perceived 5.12), who were included in all analyses reported below. There
intelligence nor intellectual engagement can be considered identi- were no significant gender differences for any of the variables
cal, or even strongly related, to intelligence as measured by ability examined in this study.
tests (correlations are typically in the range of .2–.3; e.g., Acker-
man & Heggestad, 1997; Paulhus, Lysy, & Yik, 1998). Intellectual Measures
engagement reflects motivation, interest, and enjoyment in intel-
lectual pursuits, without necessarily reflecting cognitive ability Personality. Personality was assessed using the NEO-PI-R, a
(though ability could encourage engagement, and vice versa). The standard instrument for measuring the Big Five, with excellent
items of the Ideas facet were selected to reflect intellectual en- reliability and validity (Costa & McCrae, 1992b). Cronbach’s
gagement rather than perceived intelligence (Costa & McCrae, alphas for the Big Five domains were Neuroticism ϭ .92, Extra-
1992b), and Ideas is the NEO-PI-R facet most strongly related to version ϭ .87, Openness ϭ .89, Agreeableness ϭ .91, Conscien-
measures of Typical Intellectual Engagement (r ϭ .77; Ackerman tiousness ϭ .91. Cronbach’s alphas for the six facets of Openness/
& Goff, 1994) and the similar construct Need for Cognition (r ϭ Intellect were Fantasy ϭ .84, Aesthetics ϭ .81, Feelings ϭ .67,
.78; Cacioppo, Petty, Feinstein, & Jarvis, 1996). Nonetheless, the Actions ϭ .66, Ideas ϭ .84, Values ϭ .70.
fact that both intellectual engagement and perceived intelligence
tend to be positively associated with tests of ability makes it WM. WM was assessed during fMRI using a three-back WM
important to test whether any association with Intellect is due to task. Participants were required to press one button if the item
intelligence measured as an ability. presented on the screen was identical to that presented three trials
previously, and another button if the item was different. The task
There are two possibilities for the outcome of an analysis of the was made additionally difficult by the inclusion of lure trials, in
relations among Intellect, intelligence, and brain activity support- which the stimulus matched one seen previously (and hence was
ing WM performance (assuming the correctness of our initial familiar in the context of the task) but did not match the one three
hypothesis that Intellect is associated with WM-related brain ac- back. Trial proportions were 31% targets, 19% lures, and 50%
tivity). First, any association between Intellect and WM-related nonlure/nontargets. Participants performed this task in six func-
brain activity may be eliminated by controlling for intelligence, tional scanning runs, each comprising two blocks of 32 trials
which would mean that self-rated Intellect is associated with lasting 2 s each (64 total trials per functional run). The first three
WM-related brain activity only because it reflects intelligence with trials of each block were discarded because no match was possible,
some limited degree of accuracy. Second, the association may be leaving 58 trials per run. Runs alternated between using faces and
independent of intelligence, which would indicate that the brain concrete nouns as stimuli, with order counterbalanced across par-
activity in question is probably associated with intellectual engage- ticipants; all face stimuli displayed a mildly positive (smiling)
ment, rather than with or (as well as) with intelligence. We used expression, and all nouns were emotionally neutral. Every run was
mediation tests to examine the relations among these constructs. preceded by a short video; two of these videos involved positive
As an even more specific test of the degree to which Intellect is emotion inductions, two involved negative emotion inductions,
associated with WM-related brain activity for reasons other than and two were emotionally neutral. The order of video presentation
cognitive ability, we included measures of WM capacity, in addi- was counterbalanced, and we do not focus on this variable in our
tion to tests of intelligence. analyses. (All significant effects reported in Table 1 remained
significant as main effects when controlling for stimulus type and
Method emotion condition in a series of 2 ϫ 3 repeated measures analyses
of covariance [ANCOVAs], run post hoc.) Three-back perfor-
Participants mance was assessed by the signal detection measure of accuracy,
dЈ, averaged across runs. Cronbach’s alpha for dЈ across the six
Right-handed participants were recruited from Washington Uni- runs was .84. Blocks of the three-back task in which participants
versity and the surrounding community, in St. Louis, Missouri, to omitted responses to more than 15 trials were excluded from all
participate in a neuroimaging study. (There is no overlap with the analyses. For 6 participants, three-back accuracy (and associated
sample described by Gray et al., 2003; there is substantial overlap brain activity) was therefore computed on the basis of five blocks.
with the samples described by Shamosh et al., 2008, and Fales et Accuracy was not significantly correlated with reaction times on
al., 2008.) The experimental protocol was approved by the Wash- the task, nor was reaction time correlated with any of the other
ington University Medical Center Human Subjects Committee. All variables examined in the study.
participants gave informed consent and were screened for history
of neurological or psychiatric disorders and use of psychoactive In addition, participants completed four WM span tasks outside
drugs. Participants (N ϭ 107) were selected for the present study of the scanner. These tasks required participants to keep informa-
if they had completed the NEO-PI-R and had complete imaging tion in mind despite interference (Conway et al., 2003). Two were
data. One of these participants was excluded because performance verbal and two were spatial: Operation span required keeping
on the WM task was not significantly better than chance (dЈ ϭ several words in mind over a short delay while doing math prob-
0.31, 3.19 SDs below the mean). Another participant was excluded lems. Reading span required keeping several letters in mind while
for omitting more than 20 responses in multiple blocks of the task. reading sentences out loud and judging their meaningfulness. Sym-
Finally, the participant with the highest WM score was excluded as metry span required keeping several spatial locations in mind while
an outlier (dЈ ϭ 3.80, 3.24 SDs above mean, the next highest score making symmetry judgments. Rotation span required keeping the size
being only 2.12 SDs above the mean). This left 104 participants and direction of several arrows in mind while making unrelated
judgments that required mental rotation. For all four tasks, keeping

886 DEYOUNG, SHAMOSH, GREEN, BRAVER, AND GRAY

Table 1
Correlations of Working Memory (dЈ and WMC), Intelligence (g), the Big Five, and Facets of Openness/Intellect With Brain Activity
in Four ROIs (With MNI and Talairach Coordinates)

Variable dЈ g WMC R SPC L aPFC R aPFC pMFC

MNI — .58‫ءء‬ .51‫ءء‬ 12, Ϫ69, 66 Ϫ24, 66, 6 24, 66, 18 Ϫ3, 6, 66
Talairach .58‫ءء‬ — .60‫ءء‬ 10, Ϫ70, 54 Ϫ22, 61, 8 20, 61, 19 Ϫ3, 3, 58
dЈ .51‫ءء‬ .60‫ءء‬ —
g .03 Ϫ.09 Ϫ.11 .43a‫ءء‬ .37a‫ءء‬ .34a‫ءء‬ .31a‫ءء‬
WMC .03 .06 Ϫ.04 .27‫ء‬ .21‫ء‬ .20‫ء‬ .21‫ء‬
Fantasy .00 Ϫ.06 Ϫ.09 .30‫ءء‬ .19‫ء‬ .09 .27‫ء‬
Aesthetics .05 Ϫ.01 Ϫ.04 Ϫ.06 .16 .10 .03
Feelings .23‫ء‬ .27‫ءء‬ .19‫ء‬ .01 .08 .01 Ϫ.07
Actions .23‫ء‬ .33‫ءء‬ .12 Ϫ.06 .08 Ϫ.02 .03
Ideas Ϫ.08 Ϫ.13 Ϫ.03 Ϫ.04 .12 Ϫ.01 .11
Values .01 .11 .09 .14 .21‫ء‬ .06 .27‫ءء‬
E Ϫ.15 Ϫ.10 .06 .07 .15 .10 .08
A Ϫ.05 .09 Ϫ.05 Ϫ.11 .11 Ϫ.05 Ϫ.01
C .14 .09 .01 .18 .05 .03 .15
N Ϫ.01 .02 .10 Ϫ.02
O/I .00 Ϫ.09 .01 Ϫ.09
.02 .25‫ء‬ .10 .10

Note. N ϭ 104. WMC ϭ working memory capacity; ROI ϭ region of interest; MNI ϭ Montreal Neurological Institute; R ϭ right; SPC ϭ superior parietal
cortex; L ϭ left; aPFC ϭ anterior prefrontal cortex; pMFC ϭ posterior medial frontal cortex; E ϭ Extraversion; A ϭ Agreeableness; C ϭ
Conscientiousness; N ϭ Neuroticism; O/I ϭ Openness/Intellect.
a These correlations are not independent of the initial statistical test that identified the ROIs. Their designation as significant therefore reflects that initial

test.
‫ ء‬p Ͻ .05. ‫ ءء‬p Ͻ .01.

more items in mind resulted in higher scores, on a metric ranging from tion). During each functional run, the intertrial intervals were
0.00 to 1.00. Factor analysis showed that all four tasks loaded strongly jittered across a range of 0 – 4,720 msec (0 –2 TRs) in steps of
on a single factor accounting for 70% of total variance (loadings 2,360 msec (1 TR). Each task block was preceded and followed by
ranged from .81 to .86). WM capacity (WMC) was defined as the a resting fixation block of 35 s, during which participants were
average score across all four measures (␣ ϭ .86). instructed to watch a simple dash that remained at the center of the
screen. Each scanning run began with an unanalyzed 4 TR fixation
Intelligence. Participants completed four standard measures of period that allowed the scanner to reach steady state.
intelligence: Raven’s Advanced Progressive Matrices, Set II
(Raven, Raven, & Court, 1998), the Cattell Culture Fair Intelli- fMRI Data Analysis
gence Test (Cattell, 1973), the Vocabulary subscale of the Wech-
sler Adult Intelligence Scale-Revised (Wechsler, 1997), and the Data were analyzed using Statistical Parametric Mapping 2
National Adult Reading Test-Revised (Blair & Spreen, 1989). (SPM2) software (Wellcome Department of Imaging Neuro-
Intelligence was assessed as general cognitive ability (g), com- science, 2003). Each functional run was preprocessed prior to
puted as the average of participants’ standardized scores on the analysis. Data were realigned using INRIAlign (Freire, Roche, &
four psychometric tasks (␣ ϭ .88). Common factor analysis re- Mangin, 2002) to correct for movement. Images were normalized to
vealed that a single factor explained 74% of the shared variance in Montreal Neurological Institute (MNI) stereotaxic space using a 12-
these measures, the scree plot indicated no second factor, and all parameter affine transformation followed by nonlinear warping using
variables loaded strongly and approximately equally on the first basis functions, resampled into 3-mm isotropic voxels, and smoothed
unrotated factor (range ϭ .79 –.84). Operationalizing g using using an 8-mm full-width at half-maximum Gaussian kernel.
scores derived from the first unrotated factor did not change the
results appreciably and has the disadvantage of capitalizing on For each participant, a basic contrast, task Ͼ fixation, was
sampling variability. computed across all six functional runs.3 Each 32-trial block of

fMRI Data Acquisition 3 A previous study using the same task in a completely different sample,
in order to investigate neural correlates of fluid intelligence, focused on
Whole-brain images were collected on a 3 Tesla Allegra System neural activity associated specifically with lure trials, as a phasic departure
(Siemens, Erlangen, Germany), including T1-weighted MP-RAGE from sustained activity (Gray et al., 2003). Additionally, they identified
structural images (field of view [FOV] ϭ 256 mm; 256 ϫ 256 ROIs on the basis of correlations with fluid intelligence rather than with
matrix; 1.25-mm thick axial slices) and T2 ‫ ء‬blood oxygen level- WM accuracy. These differences may explain their identification of a
dependent functional images (asymmetric spin-echo echo-planar different set of ROIs. For the present purposes, we were interested in
sequence; return time [TR] ϭ 2,360 ms; echo time [TE] ϭ 25 ms; sustained activity throughout the WM task, rather than phasic activity in
FOV ϭ 256 mm; flip angle ϭ 90°; matrix ϭ 64 ϫ 64; 4-mm thick lure trials, because the n-back task requires continual maintenance, mon-
axial slices). Each functional run comprised 149 sequential whole- itoring, updating, and integration of information, all of which are likely to
brain volumes (32 contiguous slices, 4 ϫ 4 mm in-plane resolu- be relevant to Intellect.

INTELLECT AND fMRI OF WORKING MEMORY 887

three-back performance was modeled as a boxcar function con-
volved with a canonical hemodynamic response function. The
magnitude of neural activity at each voxel was estimated using the
general linear model. This contrast produced statistical parametric
maps of the t statistic at each voxel for each participant. These
maps of the brain, indicating the difference in neural activity at
each voxel between when participants were engaged in the WM
task and when they were simply focusing on the fixation point,
were used in all subsequent analyses.

As potential neural correlates of Intellect, candidate regions of
interest (ROIs) were identified in which brain activity during the
WM task covaried with three-back accuracy between participants,
using a group-level random-effects analysis. ROIs were selected if
they comprised 15 or more contiguous voxels in which the task Ͼ
fixation contrast values correlated with dЈ at p Ͻ .001, uncorrected.
The MarsBar toolbox (Brett, Anton, Valabregue, & Poline, 2002)
was used to define these ROIs and to extract average percent signal
change values from each ROI (computed as the mean B value
across all voxels in the ROI divided by the global signal, or mean
across all voxels in the brain). Percent signal change in each ROI
was then averaged across all six functional runs (mean alpha
across all ROIs ϭ .83), and this index of brain activity was
examined for correlation with personality.

Mediation Tests Figure 1. Coronal (from the front), axial (from above), and midsagital
views of (A) neural activity associated with performing a working memory
Mediation tests indicate whether the association between two task, relative to focusing on a fixation point, for the sample as a whole
variables is due to another variable or set of variables. In an (N ϭ 104), and (B) four regions in which neural activity was associated
imaging context, mediation analyses can be used to test whether with working memory accuracy, in between-subjects analysis.
activity in a given brain region can plausibly account for the
covariation between two behavioral variables, thereby implicating We then used the t values for the contrast of task Ͼ fixation at
the region’s function in that association (Gray et al., 2003). In the each voxel across the whole brain as input for a second step of the
present study, a significant mediation by WM-related brain activity analysis that identified ROIs for which there were a positive
of the association between Intellect and WM accuracy in the correlation between activation level and between-subjects varia-
three-back task would indicate that this brain activity is responsi- tion in dЈ—in other words, brain regions in which individual
ble, at least in part, for the relation between Intellect and WM. differences in neural activity predicted task performance. We
Mediation tests were also used to assess the role of cognitive identified three ROIs on the basis of correlations with dЈ, one ROI
ability in the associations of interest. Mediation tests were com- of 62 voxels in right superior parietal cortex (SPC, Brodmann area
puted using path analysis in Amos 7.0 (Arbuckle, 2006), using 7), and ROIs of 20 voxels each in left and right lateral aPFC
maximum-likelihood estimation and the bootstrap method to test (Brodmann area 10).4 Because we had an a priori hypothesis that
the significance of indirect effects of personality on WM perfor- activity in pMFC would also be related to Intellect, and because
mance through brain activity (bootstrap N ϭ 2,000). (The boot- there were strong activations in this region for the sample as a
strap method replaces the inferior Sobel test; Shrout & Bolger, whole (see Figure 1A), we examined whether voxels in this brain
2002). Additionally, the independence of indirect effects in mul- region also showed correlations with performance that were reli-
tiple mediation analysis was tested using the SPSS multiple me- able but did not meet our stringent statistical thresholds (which
diation macro by Preacher and Hayes (2008). were used to protect against false positives in whole-brain analy-
ses). Indeed, when the statistical threshold was lowered to p Ͻ
Results .005, we identified a 21-voxel ROI in pMFC (Brodmann area 6)
superior to dorsal ACC. This ROI is shown with the other three in
Before discussing the ROIs associated with dЈ, we present an
image of the basic contrast of task Ͼ fixation (see Figure 1A), 4 Because our previous study of these data (Shamosh et al., 2008) used
thresholded at p Ͻ .001. This figure illustrates the pattern of neural a slightly different subsample, based on the availability of different mea-
activity, for the sample as a whole, when participants engaged in sures, we reported, in that article, a partially different set of ROIs based on
the WM task as opposed to when they simply focused on a fixation the same selection criteria. That study found nearly identical ROIs in right
cross. The key point is that the canonical WM brain network SPC and left aPFC, but also found four additional ROIs and did not report
(Cabeza & Nyberg, 2000; Owen et al., 2005; Wager & Smith, an ROI in right aPFC. We examined those four additional ROIs in the
2003) was engaged by the task; group-level activations were present sample and found that none of them were correlated with any
apparent in lateral PFC, parietal cortex, and regions in and adjacent NEO-PI-R variables.
to dorsal ACC.

888 DEYOUNG, SHAMOSH, GREEN, BRAVER, AND GRAY

Figure 1B, and coordinates of the point of strongest correlation one is willing to consider dЈ simply as another indicator of a trait
within each ROI are given in Table 1. Scatterplots of the correla- of WM ability, then it can be used as an additional potential
tions of dЈ with average activation in each ROI are shown in mediator, and doing so provides a particularly stringent test of
Figure 2. Importantly, the tests that identified these ROIs are whether the association of Intellect with brain activity is indepen-
statistically independent of our primary test of interest, which is dent of ability, given that association with dЈ was used to select the
the association of activation in the ROIs with facets of Openness/ ROIs in the first place.
Intellect.
Results of these mediation tests are presented in Table 2 and
Table 1 shows the correlations of the Big Five and the facets of reveal that, whereas controlling for cognitive ability did reduce the
Openness/Intellect with dЈ, intelligence, WMC, and brain activity association of Intellect with activity in left aPFC below signifi-
in the four ROIs. Confirming our hypothesis, Intellect, as repre- cance, it did not reduce the association of Intellect with activity in
sented by the Ideas facet, was significantly correlated with dЈ, g, pMFC below significance. Thus, we have evidence that the asso-
WMC, and with activity in left lateral aPFC and pMFC. Additional ciation between Intellect and activity in pMFC during a difficult
regressions showed that these associations were not moderated by cognitive task is not simply due to the fact that Intellect is asso-
gender or age. However, in one of the other two ROIs, an associ- ciated with cognitive ability.
ation with Intellect was moderated by gender (␤ for Ideas ϫ
Gender ϭ .36, p Ͻ .05), such that Intellect was associated with Neither g nor WMC could be demonstrated to mediate the
activity in right SPC, but only for women. For women, r ϭ .33, association between Intellect and activity in aPFC (dЈ did mediate
p Ͻ .05, whereas for men, r ϭ Ϫ.15, p ϭ .33. (Because this this association, but, as noted above, this is a special case because
moderated association was not predicted, we do not focus on this dЈ was used to select the ROI and was causally dependent on
ROI in additional analyses.) the brain activity in question). In contrast, WMC (and dЈ but not g)
did partially mediate the association between Intellect and activity
No other NEO-PI-R variable was associated with both cognitive in MFC. This means there was not only a significant direct path
performance and brain activity. The Values facet was significantly from Intellect to MFC but also a significant and independent
correlated with dЈ and g, but it was not correlated with WMC or indirect path, through WM ability.
any of the ROIs. The full Openness/Intellect domain score was
correlated with activity in left lateral aPFC, but it was not corre- Discussion
lated with WM or any other ROI. None of the other NEO-PI-R
traits were significantly correlated with WM, g, or activity in the As hypothesized, the Intellect aspect of the Openness/Intellect
four ROIs.5 As expected, based on the literature showing their trait domain (represented by the Ideas facet of the NEO-PI-R) was
strong behavioral and neural overlap, WM (measured by both dЈ associated both with performance on a difficult WM task and with
and WMC) and g were strongly correlated with each other, and g brain activity in two brain regions that supported accuracy in WM.
was significantly correlated with neural activity in all four ROIs. Neural activity in both left lateral aPFC and pMFC mediated the
WMC was correlated with all ROIs except the one in right lateral association of Intellect with WM, suggesting that one reason why
aPFC. people who describe themselves as more intellectual show better
WM performance is that they tend to engage these brain regions
Because the Ideas facet was significantly correlated with both dЈ more strongly during difficult cognitive operations.
and brain activity in two ROIs, we conducted a multiple mediation
test to determine whether the association between Intellect (repre- The associations of Intellect with WM performance and brain
sented by Ideas) and dЈ was due to activity in these ROIs (see activity were in the range of r ϭ .2–.3, which should be considered
Figure 3). The indirect effect of Intellect on dЈ was significant moderate, based on empirical guidelines for interpreting effect
(␤ ϭ .12, SE ϭ .047, p Ͻ .01), indicating significant mediation by sizes (Hemphill, 2003; Richard, Bond, & Stokes-Zoota, 2003).
the two ROIs. Additionally, the indirect effect through each ROI Associations of intelligence (g) and WMC with brain activity were
was significant even when controlling for the other, p Ͻ .05 for in the same range. Observe that the correlations of brain activity
both (and this remained true when controlling for both gender and with WM accuracy, during the WM task itself, were only in the
age). Thus, there are at least two independent neural mechanisms range of .3–.4 (see Table 1). Notably, neural activity in the two
by which Intellect is associated with WM accuracy, one in left PFC regions accounted for over three quarters of the variance in
lateral aPFC and one in pMFC. The extent of this mediation can be WM accuracy associated with Intellect.
quantified by noting that there was a 77% decrease in variance in
WM accuracy explained directly by Intellect, after partialing out 5 Two previous studies of different samples (Gray & Braver, 2002; Gray
activity in the two ROIs. et al., 2005) have examined the association of WM accuracy ([dЈ]) and
brain activity in the same task with personality scales measuring sensitivity
We ran additional mediation models to test whether intelligence of the behavioral inhibition and behavioral activation systems (BIS/BAS
or WM mediate the associations of Intellect with brain activity. scales; Carver & White, 1994). The first found a significant correlation of
Intelligence (g), WMC, and dЈ were all tested individually as BAS with dЈ (r ϭ .18, p Ͻ .05), when examining the first time participants
mediators. Finally, all three were used simultaneously to provide performed the task. The second replicated this finding for the first admin-
an even more stringent test of whether cognitive ability accounted istration but found that the association was not significant when averaging
for the association between Intellect and brain activity. Note that over repeated administrations of the task (r ϭ Ϫ.03). The BIS/BAS scales
there is a serious conceptual difficulty in using dЈ as a mediator were administered to the present sample; correlations of dЈ with BIS and
here, because dЈ represents performance on the task in which brain BAS were .06 and Ϫ.04, respectively, in the first block, and .08 and Ϫ.03
activity was assessed, and this performance was caused by that across all blocks (all p Ͼ .40). There were no significant correlations
brain activity. In a causal model, therefore, the arrow must run between BIS or BAS and the four ROIs.
from brain activity to dЈ, not the other way around. Nonetheless, if

INTELLECT AND fMRI OF WORKING MEMORY 889

Figure 2. Scatterplots showing the correlation of brain activity (percent signal change for task vs. fixation
averaged across all voxels in each region of interest) during the three-back working memory task with accuracy
(dЈ) in the task. R ϭ right; SPC ϭ superior parietal cortex; L ϭ left; aPFC ϭ anterior prefrontal cortex; MFC ϭ
medial frontal cortex.

We used mediation tests to determine whether the association of activity in this region. This interpretation is sensible when one
Intellect with brain activity was independent of cognitive ability, considers that Intellect subsumes perceived intelligence as well as
measured both by intelligence tests and by tests of WMC. The intellectual engagement (DeYoung et al., 2007), and to some
association of Intellect with pMFC activity was indeed indepen- limited extent perceived intelligence accurately reflects intelli-
dent of cognitive ability, whereas the association of Intellect with gence as measured by ability tests (Paulhus et al., 1998). Thus, one
left aPFC activity was not. By the standards of formal mediation plausible source of both Intellect and cognitive ability is the
tests, intelligence did not mediate the association between Intellect function of left lateral aPFC, which has been characterized as
and either brain region (and WMC only partially mediated the
association of Intellect with pMFC). Nonetheless, with g or WMC Table 2
in the model, neither Intellect nor g or WMC was significantly Tests of Mediation of the Associations Between Intellect (I) and
associated with neural activity in left aPFC. The fact that the Brain Activity by Cognitive Ability (Intelligence [g] and
effects of both Intellect and cognitive ability on left aPFC were Working Memory)
suppressed relative to their zero-order magnitude suggests that the
shared variance of Intellect and cognitive ability is associated with Paths

Figure 3. Path analysis demonstrating that working memory-related neu- Mediation model Indirect
ral activity in two brain regions (left lateral anterior prefrontal cortex and
posterior medial frontal cortex) mediates the association of Intellect (the (ROI and mediator) I to ROI I to ability Ability to ROI effect
Ideas facet of the NEO Personality Inventory-Revised) with working
memory accuracy (d Ј). Each brain region is an independent mediator at aPFC .17 .27‫ءء‬ .16 .045
p Ͻ .05. The zero-order correlation between Ideas and dЈ appears in g .18 .19‫ء‬ .16 .030
parentheses. Red paths involve neural variables, black paths do not. WM ϭ WMC .13 .23‫ءء‬ .34‫ءء‬ .077‫ءء‬
working memory. ‫ ء‬p Ͻ .05. ‫ ءء‬p Ͻ .01. dЈ .14 — .072
All 3 —
.23‫ ء‬.27‫ءء‬ .039
pMFC .22‫ ء‬.19‫ء‬ .14 .044‫ء‬
g .20‫ ء‬.23‫ءء‬ .23‫ء‬ .060‫ءء‬
WMC .20‫— ء‬ .26‫ء‬ .062
dЈ
All 3 —

Note. N ϭ 104. ROI ϭ region of interest; aPFC ϭ anterior prefrontal
cortex; WMC ϭ working memory capacity; All 3 ϭ g, WMC, and dЈ used
as simultaneous predictors; pMFC ϭ posterior medial frontal cortex.

Dashes indicate that path weights are not shown for tests using multiple

mediators.
‫ ء‬p Ͻ .05. ‫ ءء‬p Ͻ .01.

890 DEYOUNG, SHAMOSH, GREEN, BRAVER, AND GRAY

integrating information from multiple cognitive operations, across The one ROI identified that was entirely unrelated to Intellect
a variety of different cognitive tasks (Gilbert et al., 2006; Green et was in right lateral aPFC, and this ROI may not be reliable because
al., 2006; Ramnani & Owen, 2004). it did not appear when using the same technique to identify ROIs
in a largely overlapping sample (Shamosh et al., 2008). The left
In contrast to left aPFC, neural activity in pMFC remained (but not the right) aPFC ROI is consistent with previous studies of
significantly associated with Intellect even after partialing out abstract relational integration (Bunge, Helskoga, & Wendelken,
variance associated with g and WM. This suggests that intellectual 2009; Green et al., 2006) and with a review suggesting that left
engagement plays a major role in the association of Intellect with rather than right aPFC is predominantly involved in intelligence
activity in pMFC. The ROI identified in pMFC lies in a region that (Jung & Haier, 2007). Although research on WM suggests that
is involved in monitoring goal-directed performance (Ridderink- lateralization in PFC is influenced somewhat by whether visual or
hof et al., 2004) and is sensitive to the likelihood of error (Brown verbal stimuli are used (Wager & Smith, 2003), we used both
& Braver, 2005). This performance monitoring function aids in the visual and verbal stimuli in our WM task in order to identify brain
resolution of uncertainty during decisions and in resolving re- activity involved in WM regardless of modality. Our results are
sponse conflict (Brown & Braver, 2005; Ridderinkhof et al., consistent with previous research, but additional research will be
2004). Both uncertainty and response conflict should be frequent necessary to determine the significance of the left lateralization of
in the three-back WM task, as participants attempt to decide activity in aPFC in relation to complex cognition.
whether the present stimulus matches the one exactly three previ-
ously and to inhibit the urge to respond to lures. The tendency to Although Ideas was the only facet of Openness/Intellect asso-
monitor cognitive performance closely and accurately is another ciated with brain activity and performance on all cognitive tasks,
plausible source of Intellect as a trait and one that may reflect not the Values facet was associated with g and one WM performance
just intelligence but also one’s motivation and interest in intellec- variable; these associations are consistent with previous research,
tual activities (i.e., one’s intellectual engagement). reporting similar effect sizes (DeYoung et al., 2005). The finding
that the Values facet was associated with cognitive variables but
In relation to this conclusion, it is important to note that the not brain activity suggests that, compared with Ideas, Values has a
function served by pMFC is unlikely to be specific to difficult WM less direct link to the brain activity that supports WM. Nonethe-
operations or even complex cognitive tasks. This brain region is less, the behavioral associations are worth noting, given related
similarly active during much simpler cognitive tasks that involve research. Because Values reflects a liberal worldview and its
response conflict and/or uncertainty, such as Stroop tasks (Rid- alternative label is “Liberalism” (Goldberg, 1999), our finding
derinkhof et al., 2004). (Of course, the mere fact that the same may be relevant to the characterization of cognitive processes
brain region is active during different tasks does not guarantee that associated with political attitudes. They are consistent with find-
it is performing similar functions during different tasks, but many ings that liberalism, as a political orientation, is associated with
cognitive tasks clearly require the kind of performance monitoring higher IQ, relative to conservatism (Block & Block, 2006; Deary,
that this region subserves.) In order to distinguish between brain Batty, & Gale, 2008). Furthermore, one recent study reported that
activity related exclusively to WM and brain activity that would liberalism was associated with better performance and increased
support performance on simpler cognitive tasks as well, we would brain activity in pMFC, during a cognitive control task (Amodio,
have needed to include a simple cognitive task as a control con- Jost, Master, & Yee, 2007). The Values facet is not a particularly
dition. However, our concern was to detect neural correlates of the good marker of Intellect, but it is nonetheless moderately related to
personality trait Intellect, not to pin down the specific neural Intellect as well as to Openness (DeYoung et al., 2007), and it may
correlates of WM, and we were, therefore, interested broadly in be fruitful to continue investigating this politically relevant dimen-
neural processes that support cognition. Given that the variance sion of personality in future research on the cognitive and neural
responsible for most of the association between Intellect and correlates of personality.
pMFC activity appeared to be related to intellectual engagement,
rather than to cognitive ability, it seems quite likely that the A question for future research is what brain functions might
Intellect–pMFC association reflects a tendency toward monitoring contribute specifically to Openness as opposed to Intellect. This
cognitive performance that would be heightened across a wide might be revealed through fMRI by using a task more directly
variety of cognitive tasks for those high in Intellect. This brain relevant to the artistic and contemplative qualities that characterize
region may be modulated by motivation to perform any cognitive Openness. Additionally, future research on Openness and Intellect
task vigilantly and accurately. could benefit from using personality measures designed specifi-
cally to distinguish these two aspects (e.g., the Big Five Aspect
One additional region that was identified as supportive of WM Scales; DeYoung et al., 2007).
accuracy was associated with Intellect only in women. For women,
but not for men, Intellect was associated with activity in right SPC. In conclusion, Intellect and Openness are separable aspects of
We did not hypothesize any such interaction with gender. How- one larger domain of personality, which raises the question of their
ever, this brain region was strongly associated with WM perfor- discriminant validity. To our knowledge, the present study pro-
mance and is part of the canonical WM network (Cabeza & vides the first assessment of the neural correlates of either aspect.
Nyberg, 2000; Owen et al., 2005; Wager & Smith, 2003), so the Intellect subsumes traits reflecting intellectual engagement and
finding is reasonably consistent with our hypotheses. Future re- perceived intelligence. Consistent with this characterization, we
search might investigate gender differences in the relation of found that the Ideas facet of the NEO-PI-R (a good marker of
personality to cognitive and brain function, building on research Intellect) was associated with measures of intelligence and WM,
showing gender differences in brain function (e.g., Canli, Des- whereas the facets that strongly mark Openness were not. We went
mond, Zhao, & Gabrieli, 2002) and in various forms of cognitive beyond previous research (DeYoung et al., 2005) in the present
ability, including WM (e.g., Kaufman, 2007). study by investigating the biological sources of the link between

INTELLECT AND fMRI OF WORKING MEMORY 891

Intellect and WM accuracy and found two regions of PFC that capacity and its relation to general intelligence. Trends in Cognitive
mediated this association, one in the left anterior PFC, or frontal Sciences, 7, 547–552.
pole, and the other in posterior medial frontal cortex. The region in Costa, P. T., Jr., & McCrae, R. R. (1992a). Four ways five factors are basic.
pMFC, which has been implicated in performance monitoring, was Personality and Individual Differences, 13, 653– 665.
significantly associated with Intellect even after controlling for Costa, P. T., Jr., & McCrae, R. R. (1992b). NEO PI-R Professional
intelligence and WM ability. Neither region was associated with Manual. Odessa, FL: Psychological Assessment Resources.
any other Openness/Intellect facets. These findings demonstrate Deary, I. J., Batty, G. D., & Gale, C. R. (2008). Bright children become
the possibility of distinguishing two aspects of a single Big Five enlightened adults. Psychological Science, 19, 1– 6.
domain in terms of neural correlates and suggest that the functions De Pisapia, N., Slomski, J. A., & Braver, T. S. (2007). Functional special-
of pMFC may be an important substrate of Intellect that is distinct izations in lateral prefrontal cortex associated with the integration and
from cognitive ability and is perhaps driven by the motivation to segregation of information in working memory. Cerebral Cortex, 17,
engage with intellectual activities. 993–1006.
DeYoung, C. G., & Gray, J. R. (in press). Personality neuroscience:
References Explaining individual differences in affect, behavior, and cognition. In
P. J. Corr & G. Matthews (Eds.), The Cambridge handbook of person-
Ackerman, P. L., & Goff, M. (1994). Typical intellectual engagement and ality psychology. New York: Cambridge University Press.
personality: Reply to Rocklin (1994). Journal of Educational Psychol- DeYoung, C. G., Peterson, J. B., & Higgins, D. M. (2005). Sources of
ogy, 86, 150 –153. Openness/Intellect: Cognitive and neuropsychological correlates of the
fifth factor of personality. Journal of Personality, 73, 825– 858.
Ackerman, P. L., & Heggestad, E. D. (1997). Intelligence, personality, and DeYoung, C. G., Quilty, L. C., & Peterson, J. B. (2007). Between facets
interests: Evidence for overlapping traits. Psychological Bulletin, 121, and domains: Ten aspects of the Big Five. Journal of Personality and
219 –245. Social Psychology, 93, 880 – 896.
Fales, C. L., Barch, D. M., Burgess, G. C., Schaefer, A., Mennin, D. S.,
Amodio, D. M., Jost, J. T., Master, S. L., & Yee, C. M. (2007). Neuro- Braver, T. S., & Gray, J. R. (2008). Anxiety and cognitive efficiency:
cognitive correlates of liberalism and conservatism. Nature Neuro- Differential modulation of transient and sustained neural activity during
science, 10, 1246 –1247. a working memory task. Cognitive, Affective, and Behavioral Neuro-
science, 8, 239 –253.
Arbuckle, J. L. (2006). Amos 7.0 (Build 1140) [Computer software]. Freire, L., Roche, A., & Mangin, J.-F. (2002). What is the best similarity
Mount Pleasant, SC: Amos Development Corporation. measure for motion correction in fMRI? IEEE Transactions in Medical
Imaging, 21, 470 – 484.
Blair, J. R., & Spreen, O. (1989). Predicting premorbid IQ: A revision of Furnham, A., Dissou, G., Sloan, P., & Chamorro-Premuzic, T. (2007).
the national adult reading test. The Clinical Neuropsychologist, 3, 129 – Personality and intelligence in business people: A study of two person-
136. ality and two intelligence measures. Journal of Business and Psychol-
ogy, 22, 99 –109.
Block, J., & Block, J. H. (2006). Nursery school personality and political Gilbert, S. J., Spengler, S., Simons, J. S., Steele, J. D., Lawrie, S. M., Frith,
orientation two decades later. Journal of Research in Personality, 40, C. D., & Burgess, P. W. (2006). Functional specialization within rostral
734 –749. prefrontal cortex (area 10): A meta-analysis. Journal of Cognitive Neu-
roscience, 18, 932–948.
Brett, M., Anton, J.-L., Valabregue, R., & Poline, J.-B. (2002, June 2– 6). Goldberg, L. R. (1990). An alternative “description of personality:” The
Region of interest analysis using an SPM toolbox [Abstract]. Presented Big-Five factor structure. Journal of Personality and Social Psychology,
at the 8th International Conference on Functional Mapping of the Hu- 59, 1216 –1229.
man Brain, Sendai, Japan. Goldberg, L. R. (1993). The structure of phenotypic personality traits.
American Psychologist, 48, 26 –34.
Brown, J. W., & Braver, T. S. (2005, February 18). Learned predictions of Goldberg, L. R. (1999). A broad-bandwidth, public domain, personality
error likelihood in the anterior cingulate cortex. Science, 307, 1118 – inventory measuring the lower-level facets of several five-factor models.
1121. In I. Mervielde, I. Deary, F. De Fruyt, & F. Ostendorf (Eds.), Personality
psychology in Europe (Vol. 7, pp. 7–28). Tilburg, the Netherlands:
Bunge, S. A., Helskoga, E. H., & Wendelken, C. (2009). Left, but not right, Tilburg University Press.
rostrolateral prefrontal cortex meets a stringent test of the relational Gray, J. R., & Braver, T. S. (2002). Personality predicts working-memory-
integration hypothesis. NeuroImage, 46, 338 –342. related activation in the caudal anterior cingulate cortex. Cognitive,
Affective, & Behavioral Neuroscience, 2, 64 –75.
Cabeza, R., & Nyberg, L. (2000). Imaging cognition II: An empirical Gray, J. R., Burgess, G. C., Schaefer, A., Yarkoni, T., Larsen, R. J., &
review of 275 PET and fMRI studies. Journal of Cognitive Neuro- Braver, T. S. (2005). Affective personality differences in neural process-
science, 12, 1– 47. ing efficiency confirmed using fMRI. Cognitive, Affective, & Behavioral
Neuroscience, 5, 182–190.
Cacioppo, J. T., Petty, R. E., Feinstein, J. A., & Jarvis, W. B. G. (1996). Gray, J. R., Chabris, C. F., & Braver, T. S. (2003). Neural mechanisms of
Dispositional differences in cognitive motivation: The life and times of general fluid intelligence. Nature Neuroscience, 6, 316 –322.
individuals varying in need for cognition. Psychological Bulletin, 119, Gray, J. R., & Thompson, P. M. (2004). Neurobiology of intelligence:
197–253. Science and ethics. Nature Reviews Neuroscience, 5, 471– 482.
Green, A. E., Fugelsang, J. A., Kraemer, D. J., Shamosh, N. A., & Dunbar,
Canli, T., Desmond, J. E., Zhao, Z., & Gabrieli, J. D. E. (2002). Sex K. N. (2006). Frontopolar cortex mediates abstract integration in anal-
differences in the neural basis of emotional memories. Proceedings of ogy. Brain Research, 1096, 125–137.
the National Academy of Sciences, 99, 10789 –10794. Hemphill, J. F. (2003). Interpreting the magnitudes of correlation coeffi-
cients. American Psychologist, 58, 78 – 80.
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral
activation, and affective responses to impending reward and punishment:
The BIS/BAS Scales. Journal of Personality and Social Psychology, 67,
319 –333.

Cattell, R. B. (1973). Measuring intelligence with the culture fair tests.
Champaign, IL: The Institute for Personality and Ability Testing.

Christoff, K., Prabhakaran, V., Dorfman, J., Zhao, Z., Kroger, J. K.,
Holyoak, K. J., & Gabrieli, J. D. E. (2001). Rostrolateral prefrontal
cortex involvement in relational integration during reasoning. NeuroIm-
age, 14, 1136 –1149.

Conway, A. R., Kane, M. J., & Engle, R. W. (2003). Working memory

892 DEYOUNG, SHAMOSH, GREEN, BRAVER, AND GRAY

Jang, K. L., Hu, S., Livesley, W. J., Angleitner, A., Riemann, R., & of intelligence: Are they useful as proxy IQ tests? Journal of Person-
Vernon, P. A. (2002). Genetic and environmental influences on the ality, 66, 525–554.
covariance of facets defining the domains of the five-factor model of Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling
personality. Personality and Individual Differences, 33, 83–101. strategies for assessing and comparing indirect effects in multiple me-
diator models. Behavior Research Methods, 40, 879 – 891.
Jang, K. L., McCrae, R. R., Angleitner, A., Riemann, R., & Livesley, W. J. Ramnani, N., & Owen, A. M. (2004). Anterior prefrontal cortex: Insights
(1998). Heritability of facet-level traits in a cross-cultural twin sample: into function from anatomy and neuroimaging. Nature Reviews Neuro-
Support for a hierarchical model of personality. Journal of Personality science, 5, 184 –194.
and Social Psychology, 74, 1556 –1565. Raven, J., Raven, J. C., & Court, J. H. (1998). Manual for Raven progres-
sive matrices and vocabulary scales. Oxford, England: Oxford Psychol-
John, O. P., Naumann, L. P., & Soto, C. J. (2008). Paradigm shift to the ogists Press.
integrative Big Five trait taxonomy: History, measurement, and concep- Reynolds, J. R., McDermott, K. B., & Braver, T. S. (2006). A direct
tual issues. In O. P. John, R. W. Robins, & L. A. Pervin (Eds.), comparison of anterior prefrontal cortex involvement in episodic re-
Handbook of personality: Theory and research (pp. 114 –156). New trieval and integration. Cerebral Cortex, 16, 519 –528.
York: Guilford Press. Reynolds, J. R., West, R., & Braver, T. S. (2008). Distinct neural circuits
support transient and sustained processes in prospective memory and
Johnson, J. A. (1994). Clarification of factor five with the help of the working memory. Cerebral Cortex, 19, 1208 –1221.
AB5C model. European Journal of Personality, 8, 311–334. Richard, F. D., Bond, C. F., Jr., & Stokes-Zoota, J. J. (2003). One hundred
years of social psychology quantitatively described. Review of General
Jung, R. E., & Haier, R. J. (2007). The parieto-frontal integration theory Psychology, 7, 331–363.
(P-FIT) of intelligence: Converging neuroimaging evidence. Behavioral Ridderinkhof, K. R., Ullsperger, M., Crone, E. A., & Nieuwenhuis, S.
and Brain Sciences, 30, 135–187. (2004, October 15). The role of the medial frontal cortex in cognitive
control. Science, 306, 443– 447.
Kane, M. J., & Engle, R. W. (2002). The role of prefrontal cortex in Saucier, G. (1992). Openness versus intellect: Much ado about nothing?
working-memory capacity, executive attention, and general fluid intel- European Journal of Personality, 6, 381–386.
ligence: An individual-differences perspective. Psychonomic Bulletin & Shamosh, N. A., DeYoung, C. G., Green, A. E., Reis, D. L., Johnson,
Review, 9, 637– 671. M. R., Conway, A. R. A., et al. (2008). Individual differences in delay
discounting: Relation to intelligence, working memory, and anterior
Kaufman, S. B. (2007). Sex differences in mental rotation and spatial prefrontal cortex. Manuscript submitted for publication. Psychological
visualization ability: Can they be accounted for by differences in work- Science, 19, 904 –911.
ing memory capacity? Intelligence, 35, 211–223. Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and non-
experimental studies: New procedures and recommendations. Psycho-
McCrae, R. R. (1993). Openness to Experience as a basic dimension of logical Methods, 7, 422– 445.
personality. Imagination, Cognition, and Personality, 13, 39 –55. Wager, T. D., & Smith, E. E. (2003). Neuroimaging studies of working
memory: A meta-analysis. Cognitive, Affective, & Behavioral Neuro-
McCrae, R. R., & Costa, P. T. (1985). Updating Norman’s “adequate science, 3, 255–274.
taxonomy”: Intelligence and personality dimensions in natural language Wechsler, D. (1997). WAIS-III administration and scoring manual. San
and in questionnaires. Journal of Personality and Social Psychology, 49, Antonio, TX: Harcourt Brace & Company.
710 –721. Wellcome Department of Imaging Neuroscience. (2003). SPM2 [Computer
software]. Retreived June, 29, 2009, from http://www.fil.ion.ucl.ac.uk/
McCrae, R. R., & Costa, P. T. (1997). Conceptions and correlates of spm
Openness to Experience. In R. Hogan, J. Johnson, & S. Briggs (Eds.),
Handbook of personality psychology (pp. 825– 847). Boston: Academic Received March 6, 2008
Press. Revision received March 10, 2009

Moutafi, J., Furnham, A., & Crump, J. (2003). Demographic and person- Accepted April 27, 2009 Ⅲ
ality predictors of intelligence: A study using the NEO Personality
Inventory and the Myers-Brigg Type Indicator. European Journal of
Personality, 17, 79 –94.

Moutafi, J., Furnham, A., & Crump, J. (2006). What facets of openness and
conscientiousness predict fluid intelligence score? Learning and Indi-
vidual Differences, 16, 31– 42.

Owen, A. M., McMillan, K. M., Laird, A. R., & Bullmore, E. (2005).
N-back working memory paradigm: A meta-analysis of normative func-
tional neuroimaging studies. Human Brain Mapping, 25, 46 –59.

Paulhus, D. L., Lysy, D. C., & Yik, M. S. M. (1998). Self-report measures